The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill utilizes a stateful workspace where the agent's behavior is guided by the contents of files like MISSION.md, RESOURCES.md, and learning records. This creates a surface for indirect prompt injection where data in these files could override or manipulate the agent's teaching logic.
Ingestion points: The agent is instructed to read MISSION.md, RESOURCES.md, NOTES.md, and all files in the learning-records directory to calculate the "zone of proximal development" and ground its teaching.
Boundary markers: The instructions do not define clear delimiters or "ignore" warnings when the agent processes these external files.
Capability inventory: The agent has the capability to write to the file system, read from the file system, and execute CLI commands to open files.
Sanitization: There are no explicit instructions for sanitizing or validating the content extracted from workspace files before it is used to influence the agent's decision-making process.
[COMMAND_EXECUTION]: The skill explicitly instructs the agent to "open the lesson file for the user by running a CLI command." This involves invoking shell-level execution to interact with the operating system's file-handling utilities.
[DYNAMIC_EXECUTION]: The agent is tasked with creating "interactive lessons" and "assets" including "stylesheets, quiz widgets, simulators, and diagram helpers." This involves the dynamic generation and assembly of executable web content (HTML/CSS/JS) at runtime based on the subject being taught.

teach